| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Thu, 20 May 2010 16:44:33 -0500
From: Will Schmidt <will_schmidt@...t.ibm.com>
To: Thomas Gleixner <tglx@...utronix.de>
Cc: Darren Hart <dvhltc@...ibm.com>,
Jan-Bernd Themann <THEMANN@...ibm.com>, michael@...erman.id.au,
Brian King <brking@...ux.vnet.ibm.com>,
Doug Maxey <doug.maxey@...ibm.com>, dvhltc@...ux.vnet.ibm.com,
linux-kernel@...r.kernel.org, linuxppc-dev@...ts.ozlabs.org,
niv@...ux.vnet.ibm.com
Subject: Re: [PATCH RT] ehea: make receive irq handler non-threaded
(IRQF_NODELAY)
On Thu, 2010-05-20 at 16:45 +0200, Thomas Gleixner wrote:
> On Thu, 20 May 2010, Darren Hart wrote:
>
> > On 05/20/2010 01:14 AM, Thomas Gleixner wrote:
> > > On Thu, 20 May 2010, Jan-Bernd Themann wrote:
> > > > > > Thought more about that. The case at hand (ehea) is nasty:
> > > > > >
> > > > > > The driver does _NOT_ disable the rx interrupt in the card in the rx
> > > > > > interrupt handler - for whatever reason.
> > > > >
> > > > > Yeah I saw that, but I don't know why it's written that way. Perhaps
> > > > > Jan-Bernd or Doug will chime in and enlighten us? :)
> > > >
> > > > From our perspective there is no need to disable interrupts for the
> > > > RX side as the chip does not fire further interrupts until we tell
> > > > the chip to do so for a particular queue. We have multiple receive
> > >
> > > The traces tell a different story though:
> > >
> > > ehea_recv_irq_handler()
> > > napi_reschedule()
> > > eoi()
> > > ehea_poll()
> > > ...
> > > ehea_recv_irq_handler()<---------------- ???
> > > napi_reschedule()
> > > ...
> > > napi_complete()
> > >
> > > Can't tell whether you can see the same behaviour in mainline, but I
> > > don't see a reason why not.
> >
> > I was going to suggest that because these are threaded handlers, perhaps they
> > are rescheduled on a different CPU and then receive the interrupt for the
> > other CPU/queue that Jan was mentioning.
> >
> > But, the handlers are affined if I remember correctly, and we aren't running
> > with multiple receive queues. So, we're back to the same question, why are we
> > seeing another irq. It comes in before napi_complete() and therefor before the
> > ehea_reset*() block of calls which do the equivalent of re-enabling
> > interrupts.
>
> Can you slap a few trace points into that driver with a stock mainline
> kernel and verify that ?
2.6.33.4 (non-rt kernel) with similar trace_printk hooks in place...
Most data lumps look like so:
<idle>-0 [000] 1097.685337: .handle_fasteoi_irq: ENTER 260 4000
<idle>-0 [000] 1097.685339: .handle_fasteoi_irq: pre-action 260 4100
<idle>-0 [000] 1097.685339: .ehea_recv_irq_handler: ENTER c0000000e8980700
<idle>-0 [000] 1097.685340: .ehea_recv_irq_handler: napi_schedule ... c0000000e8980700
<idle>-0 [000] 1097.685341: .ehea_recv_irq_handler: napi_schedule Calling __napi_schedule ... c0000000e8980700
<idle>-0 [000] 1097.685342: .ehea_recv_irq_handler: EXIT c0000000e8980700
<idle>-0 [000] 1097.685343: .handle_fasteoi_irq: post-action 260 4100
<idle>-0 [000] 1097.685344: .handle_fasteoi_irq: EXIT. 260 4000
<idle>-0 [000] 1097.685346: .ehea_poll: ENTER c0000000e8980700
<idle>-0 [000] 1097.685352: .napi_complete: napi_complete: ENTER c0000000e8980700
<idle>-0 [000] 1097.685352: .napi_complete: napi_complete: EXIT c0000000e8980700
<idle>-0 [000] 1097.685355: .ehea_poll: EXIT !cqe rx(1) c0000000e8980700
But I did see one like this, which shows a ehea_recv_irq_handler ENTER
within a ehea_poll ENTER. (which I think is what you were expecting,
or wanted to verify..)
<idle>-0 [000] 1097.616261: .handle_fasteoi_irq: ENTER 260 4000
<idle>-0 [000] 1097.616262: .handle_fasteoi_irq: pre-action 260 4100
* <idle>-0 [000] 1097.616263: .ehea_recv_irq_handler: ENTER c0000000e8980700
<idle>-0 [000] 1097.616264: .ehea_recv_irq_handler: napi_schedule ... c0000000e8980700
<idle>-0 [000] 1097.616265: .ehea_recv_irq_handler: napi_schedule Calling __napi_schedule ... c0000000e8980700
<idle>-0 [000] 1097.616265: .ehea_recv_irq_handler: EXIT c0000000e8980700
<idle>-0 [000] 1097.616266: .handle_fasteoi_irq: post-action 260 4100
<idle>-0 [000] 1097.616268: .handle_fasteoi_irq: EXIT. 260 4000
* <idle>-0 [000] 1097.616270: .ehea_poll: ENTER c0000000e8980700
<idle>-0 [000] 1097.616282: .handle_fasteoi_irq: ENTER 260 4000
<idle>-0 [000] 1097.616283: .handle_fasteoi_irq: pre-action 260 4100
* <idle>-0 [000] 1097.616284: .ehea_recv_irq_handler: ENTER c0000000e8980700
<idle>-0 [000] 1097.616285: .ehea_recv_irq_handler: napi_schedule ... c0000000e8980700
<idle>-0 [000] 1097.616286: .ehea_recv_irq_handler: napi_schedule NOT Calling __napi_schedule... c0000000e8980700
<idle>-0 [000] 1097.616286: .ehea_recv_irq_handler: EXIT c0000000e8980700
<idle>-0 [000] 1097.616287: .handle_fasteoi_irq: post-action 260 4100
<idle>-0 [000] 1097.616289: .handle_fasteoi_irq: EXIT. 260 4000
<idle>-0 [000] 1097.616299: .napi_complete: napi_complete: ENTER c0000000e8980700
<idle>-0 [000] 1097.616300: .napi_complete: napi_complete: EXIT c0000000e8980700
<idle>-0 [000] 1097.616302: .ehea_poll: napi_reschedule COMpleted c0000000e8980700
<idle>-0 [000] 1097.616303: .napi_complete: napi_complete: ENTER c0000000e8980700
<idle>-0 [000] 1097.616304: .napi_complete: napi_complete: EXIT c0000000e8980700
* <idle>-0 [000] 1097.616306: .ehea_poll: EXIT !cqe rx(4) c0000000e8980700
Let me know if you want/need more or a variation, etc..
Thanks,
-Will
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists